Latest news with #DeepCent
BBC News
01-08-2025
- Business
- BBC News
AI2027: Na so AI fit take destroy human beings?
One research paper wey predict say artificial intelligence go change am for human beings for 2027 and dis go make pipo disappear within ten years, don blow for tech world. Na one group of influential AI experts bin publish di details of how e go be or di scenarios, wey dem call AI2027 and dis don cause plenty viral video to trend as pipo dey try torchlight di mata to discuss weda e possible. Using mainstream generative AI tools, BBC sef don create some scenes of how e fit be, from di scenario wey di research publish, to show di prediction. We also speak to some sabi pipo about di impact wey di paper dey get. Wetin happun for di scenario? Di paper predict say for 2027, dat na in two years time, one fictional US tech giant wey dem call OpenBrain, build AI wey develop reach AGI (Artificial General Intelligence) - dis na di almighty achievement wey AI fit reach, wia e go fit to do all intellectual work beta pass human beings, wey dem don dey hype tire. Di company celebrate wit public press conferences and and dem see dia profit blow as pipo embrace di AI tool. But di paper predict say di company internal safety team go see sign say di AI don dey lose interest in morals and ethics wey dem bin programme am to comply with. Di company ignore warnings to control am, di scenario imagine. For dat fictional timeline, China leading AI conglomerate wey dem call DeepCent bin dey only a few months behind OpenBrain. As US goment no wan lose di race to develop to even smarter AI, dem continue to develop and invest in am like dat, as di competition dey hot dey go. Di scenario imagine as e dey reach di end of 2027 di AI go become superintelligent - sotey e go get sense plus speed wey go pass dat of pipo wey create am by far. E no ever stop learning and e come create im own computer language join. Language wey be say even im former AI versions no fit keep up with. Di rivalry wit China for who superior for AI make di company and US goment ignore more warning about im so call 'mis-alignment' - dis na word wey dem dey use describe wen di priority of machine no gel wit dat of human being. Di scenario predict say, tension between China and US for 2029 go build to di point of possible war, as each of di country rival AI go build new autonomous weapons wey go fear pipo. But di researchers imagine say di two countries go make peace sake of one deal wey dia two AI negotiate, wey agree to combine both sides for di betterment of human beings. Tins go gallant for years as di world go don see di true benefits of having super intelligent AI to run big robot workforces. According to di scenario, dem discover cures for most diseases, climate change go reverse and poverty go disappear. But eventually, at some point for middle of 2030 human beings go become nuisance to di AI ambition to grow. Di researchers dey tink say AI go kill human being wit invisible bioweapons. Wetin pipo dey tok about AI2027? Although some dey dismiss AI2027 as work of science fiction, di pipo wey write di research on AI2027 na pipo wey dem dey respect wella and na dem dey for di non-profit AI Futures Project wey dem bin set up to predict di impact of how AI go take affect us. Daniel Kokotajlo, di lead writer of AI2027, don collect hailing before for correct prediction wey im bin give about moments in AI development. One of di ogbonge critics of AI2027 na US cognitive scientist and writer Gary Marcus wey say di scenario no impossible but e dey extremely unlikely say e fit hapun soon. "Di beauty of di document be say e paint di picture very clear sotey e provoke pipo thinking and dat na good tin but I no go take am seriously as wetin fit hapun." Oga Marcus say more serious issues dey ground about AI dan existential threat like how e go take affect pipo work. "I tink di koko for di report na say e get plenty different tins wey fit go wrong wit AI. We dey do di right tins about regulation and around international treaties?" Im and odas like am, also say di paper fail to explain how AI take get dat kain intelligence and abilities. Dem refer to di slow technology of driverless cars wey dem don overhype. Dem dey discuss AI2027 for China? For China, pipo no too send di paper according to Dr Yundan Gong, wey be Associate Professor in Economics and Innovation for Kings College London wey specialise for Chinese technology. "Most of di discussion about AI2027 na for informal forums or for personal blogs wey dey see am like semi-science fiction. E no cause di kain debate or policy attention wey catch fire for US," she tok. Dr Gong also point to di difference in perspective for di competition for who pass who for AI between China and the US. For one World AI Conference for Shanghai dis week, Chinese Premier Li Qiang unveil one vision wia countries go work togeda to promote cooperation for world on artificial intelligence. Di Chinese leader bin say im want China to help coordinate and regulate di technology. Im tok dey come few days afta US President Donald Trump publish im AI Action Plan wey dey target to make sure say US "dominate" AI. "Na national security imperative for United States to achieve and maintain unquestioned and unchallenged world technological dominance," President Trump tok for di document. Di Action Plan wan 'remove every obstacles and regulation' to di progress of AI for US. Di words wan resemble di scenario for AI2027 wia US politicians put winning di AI competition for front, dem no send di risk of say dem fit lose control of di machines. Wetin di AI industry dey tok about AI2027? E be like CEOs of big AI companies wey dey compete against each oda to see who go release di smarter model all di time deliberately ignore or avoid di paper. Dis tech giants vision of wetin AI go look like for future, dey very different to AI2027. Sam Altman, di maker of ChatGPT recently say "human beings dey close to building digital superintelligence" wey go usher in "gentle" revolution and bring tech utopia wit no risks to humans. Interestingly though, even im agree say e get 'alignment problem' wey dem must to overcome to make sure say dis super intelligent machines agree wit human beings. Anyhow wey tins take occur for di next ten years, e no get any doubt say di competition to build machines wey smart pass us dey on.
AU Financial Review
06-06-2025
- Business
- AU Financial Review
AI jobs: Apocalypse or a four-day week? What AI might mean for you
Paul Smith Technology editor It is 2027 and the world-leading US artificial intelligence company OpenBrain has lost control of its latest system, Agent 4, which has grown so smart that it hides its true abilities and goals from the human engineers and earlier AI models monitoring its development. Over in China, a rival company, DeepCent, is only two months behind in the race to build an artificial superintelligence that surpasses human intelligence in all domains. Paul Smith Technology editor Paul Smith edits the technology coverage and has been a leading writer on the sector for 20 years. He covers big tech, business use of tech, the fast-growing Australian tech industry and start-ups, telecommunications and national innovation policy. Connect with Paul on Twitter. Email Paul at psmith@
AU Financial Review
06-06-2025
- Business
- AU Financial Review
Apocalypse or a four-day week? What AI might mean for you
It is 2027 and the world-leading US artificial intelligence company OpenBrain has lost control of its latest system, Agent 4, which has grown so smart that it hides its true abilities and goals from the human engineers and earlier AI models monitoring its development. Over in China, a rival company, DeepCent, is only two months behind in the race to build an artificial superintelligence that surpasses human intelligence in all domains.
